Celecoxib Induces Apoptosis by Inhibiting 3-Phosphoinositide-dependent Protein Kinase-1 Activity in the Human Colon Cancer HT-29 Cell Line*

Nonsteroidal anti-inflammatory drugs, which inhibit cyclooxygenase (COX) activity, are powerful antineoplastic agents that exert their antiproliferative and proapoptotic effects on cancer cells by COX-dependent and/or COX-independent pathways. Celecoxib, a COX-2-specific inhibitor, has been shown to reduce the number of adenomatous colorectal polyps in patients with familial adenomatous polyposis. Here, we show that celecoxib induces apoptosis in the colon cancer cell line HT-29 by inhibiting the 3-phosphoinositide-dependent kinase 1 (PDK1) activity. This effect was correlated with inhibition of the phosphorylation of the PDK1 downstream substrate Akt/protein kinase B (PKB) on two regulatory sites, Thr308 and Ser473. However, expression of a constitutive active form of Akt/PKB (myristoylated PKB) has a low protective effect toward celecoxib-induced cell death. In contrast, overexpression of constitutive active mutant of PDK1 (PDK1A280V) was as potent as the pancaspase inhibitor, benzyloxycarbonyl-Val-Ala-Asp-fluoromethylketone, to impair celecoxib-induced apoptosis. By contrast, cells expressing a kinase-defective mutant of PDK1 (PDK1K114G) remained sensitive to celecoxib. Furthermore, in vitro measurement reveals that celecoxib was a potential inhibitor of PDK1 activity with an IC50 = 3.5 μm. These data indicate that inhibition of PDK1 signaling is involved in the proapoptotic effect of celecoxib in HT-29 cells.

lesser extent, the risk of breast (3), esophagus (4), prostate (5), and stomach (6) cancers. Although the precise mechanisms for the chemopreventive effects of NSAIDs are not yet known, the ability of these drugs to induce inhibition of cell proliferation, potentiation of immune response, inhibition of angiogenesis, and induction of apoptosis has been reported in recent years (7,8). The most characterized target for NSAIDs is cyclooxygenase (COX), which catalyzes the synthesis of prostaglandins from arachidonic acid (9). There are two known COX isoforms, COX-1 and COX-2, with distinct expression patterns and biological activities (1,7). COX-1 is a constitutively expressed enzyme found in most tissues and remains unaltered in colorectal cancer, while COX-2 expression can be up-regulated by a variety of cytokines, hormones, phorbol esters, and oncogenes in colorectal adenomas and adenocarcinomas (10). The molecular basis of the chemopreventive effects of NSAIDs for colon cancer has been attributed mainly to inhibition of COX-2 by induction of the susceptibility of cancer cells to apoptosis (11). Consistent with this, null mutation of COX-2 in APC ⌬716 knockout mice, a murine model of familial adenomatous polyposis, restored apoptosis and reduced the size and the number of colorectal adenomas (12). Similar regression of adenomas has been observed by treatment of Min mouse with the NSAID sulindac (13). However, observations relating to the proapoptotic effect of NSAIDs lead to contradictory conclusions and demonstrate that they act via COX-dependent and COX-independent mechanisms (for a review, see Ref. 11). For example, the addition of exogenous prostaglandins to a colon cancer cell line that lacks COX activity cannot reverse the proapoptotic effect of sulindac sulfide, a metabolite derived from sulindac (14). In the same way, sulindac sulfone, another sulindac metabolite that does not inhibit COXs, affects tumor growth in animal models (15) and induces apoptosis in cultured cancer cells expressing or not expressing COXs (16,17). These results suggest that molecular targets of NSAIDs other than COXs might exist and thereby would provide a link between the chemoprotective effect of NSAIDs on cancer cells and their level of COX expression. Recent studies have identified a series of new molecular targets for NSAIDs mainly involved in signaling pathways including 15-lipoxygenase-1 (18), peroxisome proliferator-activated receptors (19), extracellular signal-regulated kinase 1/2 signaling (20), NF-B (21), p70S6 kinase (22), p21 ras signaling (23), and Akt/PKB kinase (5).
Here, we report the first evidence that celecoxib potently induces apoptosis in the colon cancer cell line HT-29, which lacks COX-2 catalytic activity (29) and specifically inhibits the 3-phosphoinositide-dependent kinase PDK1 activity, the Akt/ PKB upstream kinase. These data suggest that celecoxib induces apoptosis through a target other than COX-2 by controlling the major antiapoptotic PDK1/Akt/PKB pathway.
[␥-32 P]ATP (specific activity 3000 Ci/mmol) was from PerkinElmer Life Sciences. IL-13 and cDNA encoding for the myristoylated and dead forms of HA-tagged Akt/PKB were kindly provided by Dr. A. Minty (Sanofi Elf Biorecherche), Dr. T. F. Franke (Columbia University, New York), and Dr. P. N. Tsichlis (Fox Chase Cancer Center, Philadelphia, PA), respectively. cDNA encoding for wild-type PDK1, constitutively active form PDK1 A280V , and kinase-defective PDK1 K114G were a generous gift from Dr. F. Liu (University of Texas Health Science Center, San Antonio, TX). The enhanced chemiluminescence detection kit was from Amersham Biosciences.
Cell Culture-HT-29 cells were cultured as previously described (30). 1 day after seeding, media were changed, and cells were treated with different concentrations of celecoxib (50 -100 M). An equivalent amount of the carrier (Me 2 SO) was added to untreated cells. In order to eliminate a putative interaction between G418 antibiotic and celecoxib, stable transfected cells were cultured in the absence of G418 during the drug treatment. When required, celecoxib at 50 -100 M and IL-13 (30 ng/ml) were added from 4 to 24 h.
Stable Transfection-The expression constructs (pCMV6/HA-tagged Myr PKB, pcDNA3.0/His-tagged WT PDK1, pcDNA3.0/His-tagged PDK1 A280V , and pcDNA3.0/His-tagged PDK1 K114G ) and control pCMV6 and pcDNA3.0 vectors were introduced into exponential growing HT-29 cells using the Superfect TM transfection kit. Transfected cells were cultured in complete medium for 48 h and then selected for 3 weeks in a medium containing 800 g/ml G418. Finally, G418-resistant cells were routinely maintained in a medium containing 250 g of G418. Expression levels of each constructs were determined by Western blot analysis using anti-HA and anti-His antibodies.
Cell Viability-Cells were seeded for 24 h onto 24-well plates at 1 ϫ 10 6 cells/plate and then were exposed to various concentrations of celecoxib for different times. When used, 100 M caspase inhibitor zVAD-fmk (Cliniscience S.A.) was added 1 h before celecoxib. During the treatment, the percentage of cells floating in the medium increased over time. Adherent cells were rinsed three times with phosphatebuffered saline and were harvested by trypsinization. At each time of celecoxib treatment, floating cells were recovered by centrifugation of medium at 3200 ϫ g for 5 min. Both adherent and floating cells were combined for the assessment of cell viability, which was determined by trypan blue exclusion. All data points shown are mean values Ϯ S.E. of four independent experiments. The data points from triplicates of an individual experiment were averaged, and the data points shown are the mean of these averages from four experiments. Statistical calculation was done using Student's t test.
DNA Fragmentation Analysis-HT-29 cells were cultured in the presence or absence of 100 M celecoxib for various times. Floating and adherent cells were collected, rinsed twice with phosphate-buffered saline, and resuspended in lysis buffer containing 1% Nonidet P-40, 2 mM EDTA, and 50 mM Tris (pH 7.5) for 1 h. After centrifugation, the supernatants were treated with 5 g/ml of ribonuclease A at 37°C for 1 h, and then proteinase K was added at 2.5 g/ml for 2 h at 37°C. DNA was precipitated by 75% ethanol and 3 M sodium acetate at Ϫ80°C for 2 h, and pellet was resuspended in TE buffer. Each sample was electrophoresed on a 1.6% agarose gel and visualized by ethidium bromide staining.
Flow Cytometry Analysis-In all experiments, floating and freshly trypsinized attached cells were pooled. Thereafter, apoptotic, necrotic, and damaged cells were separated by flow cytometry. Samples were then run on a FACSCalibur flow cytometer (San Jose, CA) equipped with an argon laser and filter configuration for fluorescein isothiocyanate (FITC)/propidium iodide (PI) dye combination. Light scatter and fluorescence signals were subjected to linear and logarithmic amplification, respectively. At least 10,000 events were acquired and analyzed with CellQuest software. The quantitative determination of the percentage of cells undergoing apoptosis was performed using an annexin V-FITC apoptosis detection kit (Cliniscience S.A.) according to the manufacturer's instructions. In brief, 5 ϫ 10 5 washed cells resuspended in 100 l of annexin V binding buffer were simultaneously incubated with 5 l of FITC-conjugated annexin V and 5 l of PI for 15 min at room temperature. Before cytometric analysis, the cell suspension was supplemented with 500 l of annexin V binding buffer.
Electron Microscopy-Transmission electron microscopy was performed on cells treated or not with 100 M celecoxib for 24 h. Samples were embedded in Epon as previously reported (31). Ultrathin sections were examined in a Jeol JEM-100 CX11 electron microscope operated at 100 kV.
In Vitro PDK1 Immunoprecipitation Kinase Assay-The assay was carried out in two stages. In the first stage, active PDK1 was immunoprecipitated from HT-29 cells transfected with the empty vector, WT PDK1, PDK1 A280V , or PDK1 K114G constructs. Cells were treated or not with 100 M celecoxib for 24 h and lysed in buffer A containing 50 mM Tris-HCl (pH 7.5), 0.1% Triton X-100, 1 mM EDTA, 1 mM EGTA, 50 mM sodium fluoride, 10 mM sodium ␤-glycerophosphate, 1 mM activated sodium orthovanadate, 0.1% (v/v) 2-mercaptoethanol, and 1 M microcystin at 4°C for 1 h. Samples were centrifuged at 5000 ϫ g for 10 min, and supernatants were precleared with 50% protein G-agarose beads diluted in buffer A. In the same time, 4 g of anti-PDK1 or normal sheep IgG were incubated with 100 l of 50% protein G-agarose beads at 4°C for 1 h. One mg of each precleared cell lysate was incubated for 2 h with protein G-agarose beads bearing anti-PDK1 on a rotator at 4°C to immunoprecipitate active PDK1. In the second stage, inactive serum-and glucocorticoid-regulated kinase (SGK) was incubated with immunoprecipitated PDK1 and Mg 2ϩ /ATP to activate the SGK before the addition of [␥-32 P]ATP and Akt/SGK-specific substrate peptide (RPRAATF). The activated SGK used Mg 2ϩ /[␥-32 P]ATP to phosphorylate the Akt-specific substrate peptide. Immunoprecipitates were preincubated with inactive SGK enzyme (500 ng) for 30 min at 30°C and incubated with 66 M Akt/SGK substrate peptide in a shaking incubator for another 30 min, and 1 Ci/ml of [␥-32 P]ATP was added to start the reaction. The phosphorylated peptide substrate was then separated from the residual [␥-32 P]ATP using P81 phosphocellulose paper and quantitated by using a scintillation counter after three washes with 0.75% phosphoric acid and two washes with acetone. Values are from three separate experiments.
In Vitro Akt/PKB Immunoprecipitation Kinase Assay-HT-29 cells were treated or not with 100 M celecoxib for different times in the presence or absence of 30 ng/ml IL-13 and then lysed at 4°C for 1 h in buffer A containing 50 mM Tris-HCl (pH 7.5); 1% Triton X-100; 1 mM EDTA; 1 mM EGTA; 50 mM sodium fluoride; 10 mM sodium ␤-glycerophosphate; 0.1% (v/v) 2-mercaptoethanol; 0.1 mM phenylmethylsulfonyl fluoride; and 1 g/ml aprotinin, pepstatin, and leupeptin. Samples were centrifuged at 5000 ϫ g for 10 min, and supernatants were precleared with 50% protein G-agarose beads diluted in buffer A. In the same time, 4 g of anti-Akt/PKB or normal rabbit IgG were incubated with 100 l of 50% protein G-agarose beads at 4°C for 90 min. 1 mg of each precleared cell lysates was incubated for 2 h with protein G-agarose beads bearing anti-Akt/PKB on a rotator at 4°C to immunoprecipitate Akt/PKB. This precipitate was next used to phosphorylate a specific substrate, the recombinant murine BAD protein expressed in Escherichia coli. Briefly, 3 g of recombinant BAD were incubated with Akt/PKB-antibody-protein G-agarose complexes in the presence of magnesium/ATP mixture for 10 min at 37°C. After three washes with buffer A, samples were boiled for 5 min, resolved by 10% SDS-PAGE, and transferred onto nitrocellulose membrane. The membranes were blocked with 5% nonfat dry milk in TBST and incubated with a 1:500 dilution of anti-phospho-BAD (Ser 136 ) for 2 h at room temperature. Primary antibody was revealed using a goat anti-rabbit horseradish peroxidase-conjugated IgG for 1 h at room temperature. After revelation, the Akt/PKB activity was determined after gel scanning.

Celecoxib Induces Apoptosis in the Colon Cancer Cell Line
HT-29 -Since evidence indicates that celecoxib induced apoptosis and inhibited cell cycle progression in colorectal carcinoma cells in culture by mechanisms independent of COX-2 inhibition (26), we have examined the dose-and time-dependent effects of celecoxib on the viability of HT-29 cells, a cell line that lacks COX-2 activity (29). As shown in Fig. 1A, celecoxib significantly induced the cell death in a time-and dose-dependent manner. Treatment with 100 M celecoxib for 24 h caused a loss of viability up to 65%. This effect has been previously observed in many cell lines including prostate cancer PC3 cells (5), lung carcinoma LLC cells (26), and colon cancer HCT-15 and HCT-116 cells (26).
In order to determine whether the cytotoxic effect of celecoxib was due to apoptosis, HT-29 cells were treated with increasing concentrations of this drug in the presence of the large spectrum caspase inhibitor zVAD-fmk (Fig. 1B). In the presence of zVAD-fmk, a reduction in cell death was observed, whatever the concentrations of celecoxib used. Celecoxib-dependent apoptosis was also confirmed by detection of oligonucleosomal cleavage of DNA in HT-29 cells after treatment with 100 M celecoxib for 16 and 24 h (Fig. 2A). However, as depicted in Fig. 1B, a portion of HT-29 cells was always sensitive to celecoxib despite the presence of zVAD-fmk inhibitor, suggesting that a caspase-independent cell death may also occur in celecoxib-treated HT-29 cells. A better understanding of this caspase-independent cell death induced by celecoxib will require further investigations. Examination of the celecoxibtreated cells by electron microscopy revealed morphological alterations of apoptosis after being treated with 100 M celecoxib for 24 h including chromatin condensation (Fig. 2B). In addition, the externalization of phosphatidylserine, an early event of apoptotic process, was analyzed by flow cytometry with the annexin V-binding assay after treatment of HT-29 cells with 100 M celecoxib for the indicated times (Fig. 2C). Apoptotic (annexin V ϩ /PI Ϫ ), necrotic (annexin V ϩ /PI ϩ ), and viable cells (annexin V Ϫ /PI Ϫ ) were separated on the basis of a double labeling for annexin V-FITC and PI, a membrane DNA stain. Treatment with 100 M celecoxib caused a significant increase of apoptotic cells (annexin V ϩ /PI Ϫ ) with 12, 20, and 29% for 8, 16, and 24 h of treatment, respectively. In parallel, necrotic cells (annexin V ϩ /PI ϩ ) were detected, whatever the time of treatment, to reach 37% after 24 h of treatment (Fig. 2C). This latter effect is consistent with the secondary necrosis process, which usually comes after apoptosis in cells growing in culture.
In contrast to several reports that linked COX-2 inhibitorinduced apoptosis to Bcl-2 down-regulation (32, 33), we demonstrated by Western blot analysis that celecoxib did not affect Bcl-2 expression in HT-29 cells after 24 h of treatment (Fig.  2D). Furthermore, COX-2 expression level was unaffected throughout the time of celecoxib treatment (Fig. 2D).
Recent studies have reported that celecoxib induced apoptosis by inhibiting Akt/PKB activity in prostate cancer cells (5). Since celecoxib caused programmed cell death in HT-29 cells that lack COX-2 activity (29), we checked whether the observed cytotoxic effect of celecoxib was mediated by a modulation of Akt/PKB activity.
Celecoxib Reduces the Phosphorylation State of Akt/PKB-Akt/PKB, the major downstream effector of PI 3-kinase, is a Ser/Thr protein kinase that has been shown to exert a crucial role in the regulation of several cellular signaling pathways (reviewed in Ref. 34). By stimulation of PI 3-kinase with growth factors and cytokines, Akt/PKB is recruited from the cytosol to the plasma membrane via its pleckstrin homology domain and is phosphorylated at two regulatory sites, Thr 308 and Ser 473 , essential for its activation (34). Activated Akt/PKB, that has been involved in protecting cells from apoptosis, can phosphorylate BAD (35), caspase 9 (36), IkB kinase (37), glycogen synthase kinase-3␤ (38), and forkhead transcription factors (39), leading to their inactivation and to cell survival.
To evaluate the effect of celecoxib on Akt/PKB activity, HT-29 cells were treated with 100 M celecoxib for 24 h, and phosphorylation at Thr 308 and Ser 473 was examined using specific phosphoantibodies (Fig. 3A). Despite the slight basal phosphorylation level of Akt/PKB in the control condition, celecoxib significantly reduced Akt/PKB phosphorylation at Thr 308 and Ser 473 . The effect of celecoxib was confirmed in two independent ways: first, by measuring the Akt/PKB-mediated phosphorylation level of GSK-3␤, a direct in vivo downstream substrate of Akt/PKB (40) (Fig. 3A), and second, by determining kinase activity of Akt/PKB after immunoprecipitation (Fig. 3B). The loss of Akt/PKB phosphorylation mediated by celecoxib was correlated with an equivalent decrease of GSK-3␤ phosphorylation and an inhibition of its activity.
In order to confirm the inhibitory effect of celecoxib, we increased Akt/PKB phosphorylation by using IL-13, a pleiotropic cytokine that is known to activate the PI 3-kinase/Akt/PKB pathway in HT-29 cells (41, 42). As expected, IL-13 increased the phosphorylation level of Akt/PKB at the two major sites ( Fig. 3A) and its activity (Fig. 3, A (phospho-GSK-3␤ panel) and B) when compared with untreated HT-29 cells. However, celecoxib impaired the IL-13-mediated phosphorylation and kinase activity of Akt/PKB (Fig. 3, A and B). This effect was not caused by the attenuation of PI 3-kinase activity because celecoxib did not display any appreciable inhibition of PI 3-kinase immunoprecipitated from HT-29 cell lysates (data not shown).
To determine whether celecoxib-induced apoptosis was mediated through the inhibition of Akt/PKB, we next investigated the effect of celecoxib on HT-29 cells overexpressing a constitutively active form of Akt/PKB ( Myr PKB) bearing a myristoylated N terminus motif, sufficient to trigger its plasma membrane recruitment and to increase its activity (43). As shown in Fig. 3C, overexpression of Myr PKB detected by Western blot analysis using both anti-HA and anti-Akt/PKB antibodies was correlated with an increase of GSK-3␤ phosphorylation. Surprisingly, overexpression of Myr PKB has only a slight protective effect toward celecoxib-induced cell death (Fig. 3D). Only a 10 -15% reduction in cell death was observed. Taken together, these results suggest that celecoxib-induced apoptosis is independent of PI 3-kinase activity and that Akt/PKB may contribute only in part to the cytotoxic effect of celecoxib.
Overexpression of Constitutively Active PDK1 Protects HT-29 Cells from Celecoxib-induced Apoptosis-According to the current model of Akt/PKB activation, the enzyme undergoes cytokine-stimulated translocation to the plasma membrane by the binding of PI 3-kinase products, leading to its phosphorylation on Thr 308 by the membrane-associated upstream kinase PDK1 (44,45). Phosphorylation on Ser 473 should involve PDK1 together with a fragment of the C terminus of PRK2 termed PIF (PDK1-interacting fragment) (46). To explore whether PDK1 was involved in the celecoxib-mediated effect on cell death induction, His-tagged WT PDK1, constitutively active PDK1 A280V , and kinase-defective PDK1 K114G constructs (47) were introduced in HT-29 cells, and their respective expression levels were detected using an anti-His tag antibody (Fig. 4A). The effect of celecoxib on the state of Akt/PKB phosphorylation was measured in HT-29 cells overexpressing the different PDK1 constructs. As shown in Fig. 4B, overexpression of WT PDK1 enhanced by 25% the Thr 308 phosphorylation of Akt/ PKB associated with a proportional increase of GSK-3␤ phosphorylation (Fig. 4C), when compared with control cells (see Fig. 3A). As in control cells, celecoxib greatly reduced Akt/PKB and GSK-3␤ phosphorylation states. Overexpression of constitutive active PDK1 A280V form was shown to potentate the Thr 308 phosphorylation of Akt/PKB and consequently the phosphorylation of GSK-3␤. Despite the presence of celecoxib, a high phosphorylation level was maintained on both Akt/PKB and GSK-3␤ kinases, suggesting that the constitutive active PDK1 A280V form was able to turn off the inhibitory effect of celecoxib on Akt/PKB phosphorylation. Finally, a similar reduction of phosphorylation of Akt/PKB and GSK-3␤ was observed in celecoxib-treated cells overexpressing the kinasedefective PDK1 K114G (Fig. 4, B and C) than in celecoxib-treated control cells (see Fig. 3A). Each transfected cell population was next treated with 50 -100 M celecoxib for different times, and the cell viability was determined (Fig. 5). Overexpression of WT PDK1 and constitutively active PDK1 A280V allowed us to protect the cells from celecoxib-induced apoptosis by 30 and 65% with 100 M celecoxib as compared with cells expressing the When required, 30 ng/ml IL-13 was added throughout the celecoxib treatment. Cell lysates were analyzed for Akt/PKB phosphorylation at threonine 308 (top panel, phospho-Akt/PKB Thr 308 ) and serine 473 (second panel, phospho-Akt/PKB Ser 473 ) using phosphospecific antibodies. The expression of total Akt/PKB was detected using an anti-pan-Akt/PKB antibody (third panel, Akt/PKB). The same cell lysates were submitted to Western blot analysis using a phosphospecific antibody directed against GSK-3␤ (fourth panel, phospho-GSK-3␤). Total expression of GSK-3␤ was determined by using an anti-pan-GSK-3␤ antibody (bottom panel, GSK-3␤). Densitometry was performed on the original blots, and the percentage of maximum phosphorylation was depicted. B, control and IL-13-stimulated cells were treated or not with 100 M celecoxib for the indicated times. Akt/PKB activity was measured after immunoprecipitation using specific antibody and incubation with recombinant BAD protein as detailed under "Experimental Procedures." Data are the mean Ϯ S.E. of three separate experiments. C, HT-29 cells were transfected with pCMV6 vector containing ( Myr PKB) or not containing (vector) the myristoylated-constitutive active form of Akt/PKB. The expression level of Myr PKB was evaluated by Western blot analysis using an anti-HA antibody (top panel) and pan-Akt/PKB antibody (second panel). Note that this latter antibody was able to recognize both endogenous and HA-tagged Akt/PKB. The activity level of Myr PKB was measured by Western blot analysis using a phosphospecific GSK-3␤ antibody (third panel, phospho-GSK-3␤) by comparison with a pan-GSK-3␤ antibody (bottom panel). D, stably transfected HT-29 cells overexpressing the empty vector (vector) or the HA-tagged constitutive active form of Akt/PKB ( Myr PKB) were treated for various times with different concentrations of celecoxib. Cell viability was measured as detailed in the legend to Fig. 1.  FIG. 5. Activated PDK1 protects HT-29 cells from celecoxib- empty vector, respectively. By contrast, celecoxib treatment of HT-29 cells overexpressing kinase-defective PDK1 K114G induced the same apoptotic behavior as celecoxib-treated cells expressing the empty vector. These results demonstrate that the constitutively active PDK1 A280V mutant abrogates celecoxib-mediated Akt/PKB dephosphorylation and protects HT-29 cells from celecoxib-induced apoptosis, suggesting that PDK1 could play a crucial role in celecoxib-induced apoptosis.
Celecoxib Inhibits PDK1 Activity-The Ser/Thr kinase activity of PDK1 was measured in immunoprecipitates derived from HT-29 cells expressing either the empty vector, WT PDK1, constitutive active PDK1 A280V , or kinase-defective PDK1 K114G mutants in the presence or absence of celecoxib. Overexpression of WT PDK1 or constitutive active PDK1 A280V increased kinase activity by 2.5-and 5-fold, respectively (Fig. 6A). Expression of kinase-defective PDK1 K114G did not affect endogenous PDK1 activity, confirming that the mutant does not have a dominant inhibitory effect on the Ser/Thr kinase activity (48). Cell treatment with celecoxib markedly inhibited Ser/Thr kinase activity in control cells and in cells overexpressing WT PDK1 or kinase-defective PDK1 K114G mutant. By contrast, celecoxib did not greatly affect Ser/Thr kinase activity in cells overexpressing the constitutive active form of PDK1. In addi-tion, the time course of the inhibition of PDK1 (assayed by the Akt/PKB activity) was correlated with that of the induction of apoptosis (Fig. 6B). In order to ascertain that the activity of PDK1 is sensitive to celecoxib, anti-PDK1 immunoprecipitates derived from HT-29 cells were incubated with concentrations of celecoxib ranging from 0.5 to 50 M (Fig. 6C). Celecoxib inhibited in a dose-dependent manner Ser/Thr kinase activity with an IC 50 ϭ 3.5 M. Thus, celecoxib-induced apoptosis appears to be mediated by a selective inhibitory effect of celecoxib on the Ser/Thr kinase activity of PDK1. DISCUSSION Celecoxib belongs to the new generation of NSAIDs that selectively inhibits COX-2 activity and has been shown to decrease polyp number and size in the Apc mutant Min mouse model of adenomatous polyposis (28). In clinical trials in familial adenomatous polyposis patients, celecoxib efficacy reduced the formation of colonic polyps without gut toxicity (25). Furthermore, celecoxib effectively inhibited tumor growth and enhanced survival in the mouse model of urinary bladder cancers (49), suggesting that it could be relevant to chemoprevention of cancer types other than colon cancers. However, it remains unclear how celecoxib exerts its preventive effect on tumor growth and what is the relative importance of COX-2 inhibition on the arrest of proliferation and the triggering of apoptosis in cancer cells in in vivo and in vitro contexts.
Several lines of evidence indicate that induction of apoptosis by celecoxib occurs in vitro by mechanisms independent of COX-2 inhibition (26). In the present study, we demonstrate that celecoxib induces morphological changes characteristic of apoptosis in HT-29 cells despite the fact that this cell population lacks COX-2 activity (29). This result is in agreement with those recently published demonstrating that celecoxib caused similar cytotoxic effects in colorectal cells irrespective of COX-2 expression and in mouse embryo fibroblasts derived from Cox-2 heterozygous or Cox-2 null C57/BL6 mice (26). Moreover, celecoxib displays a time-and dose-dependent efficacy higher than that of other NSAIDs and raises the possibility that it induces apoptosis through targets other than COX-2.
Here we show that celecoxib inhibits PDK1, thereby reducing Akt/PKB phosphorylation at Thr 308 and Ser 473 . The current model of Akt/PKB activation suggests that its initial phosphorylation at Thr 308 is mediated by the membrane-associated upstream kinase PDK1 and is followed by phosphorylation on Ser 473 , which occurs through an autophosphorylation mechanism (50) or PDK1 together with the PIF fragment of PRK2 (46). Our results show that in HT-29 cells, overexpression of constitutive active PDK1 A280V protects cells from celecoxibinduced cell death. This protection is abolished in cells overexpressing the kinase-defective PDK1 K114G mutant. In addition, celecoxib directly inhibits PDK1 activity in a Ser/Thr kinase immunoprecipitation assay. These results demonstrate that the inhibition of Akt/PKB phosphorylation and the induction of apoptosis by celecoxib are mainly due to the inhibition of the Akt/PKB upstream kinase PDK1. Accordingly, overexpression of the constitutive active form of Akt/PKB ( Myr PKB) does not efficiently protect cells from celecoxib-induced apoptosis, suggesting that the antiproliferative effect of celecoxib is not solely mediated through an inhibition of Akt/PKB. As PDK1 is known to phosphorylate members of the AGC family other than Akt/ PKB including p70 S6 kinase (51), p90 RS6 kinase (52), and protein kinase C-related kinases (53), their respective roles in celecoxib-induced apoptosis warrant further investigations. Recently, Harada et al. (54) have shown that PDK1 downstream substrate p70 S6 kinase is responsible for site-specific phosphorylation of BAD, inducing the inactivation of this proapoptotic molecule and cell survival. Studies are in progress in our laboratory to investigate whether celecoxib could induce apoptosis by interfering with the p70 S6 kinase-dependent phosphorylation of BAD. However, other mechanisms cannot be excluded because celecoxib has been shown to interfere with other apoptosis signaling such as intracellular Ca 2ϩ increase and ERK2 inhibition as shown in prostate cancer cell lines (55).
At the cellular level, celecoxib induces an G 0 /G 1 cell cycle arrest in HT-29 cells prior to the induction of apoptosis (56). This effect is reminiscent to that obtained when PDK1 activity is inhibited by PDK1 antisense oligonucleotides (53). However, celecoxib neither perturbs the organization of normal intestinal epithelium nor induces apoptosis in enterocytes of celecoxibtreated mice (26). Further studies are needed to investigate whether the PDK1 signaling pathway of colon carcinoma cells is more sensitive to celecoxib-induced inhibition. Recent antisense oligonucleotide approaches suggest that PDK1 can have distinct roles in different cell types (53,57). Alterations in the PI 3-kinase/PDK1/PKB pathway are frequently associated with the pathogenesis of colorectal carcinomas (58); thus, it would be of interest to identify the effector of PDK1 responsible for the induction of cell death in colon cancer cells. Finally, besides the fact that celecoxib certainly represents a future therapeutic option for the treatment of human cancer, it could be useful for the discovery of new molecules having a wealth of inhibitory capacities on the PDK1/Akt/PKB signaling pathway.